- Market
- Saudi Arabia
- Written for
- Head of Digital, CTO
- Languages
- EN · AR
The method
Four steps from "the site feels slow" to a speed that stays
The order matters: measurement before diagnosis, diagnosis before fixes, and a guard before the engagement ends — because a speed nobody protects is borrowed, not owned.
Measure in the field
Real-visitor timings are collected and segmented — by page, by device class, by network — so the question becomes specific: which pages are slow, for whom, and at which moment of loading. A ranked list of slow pages that actually carry revenue.
Trace each delay to its cause
For each slow page, the waterfall is read until the delay has a name — an unoptimised image, a blocking third-party tag, a heavy script bundle, a sluggish backend response — rather than a general accusation against "the site". Causes, not symptoms, each tied to the pages it slows.
Fix in order of commercial impact
Fixes are sequenced by what they return, not by what is technically interesting: a delay on the enquiry form or the checkout outranks a delay on a page nobody transacts from. The scarce engineering time spent where the leak is.
Install the budget
Size and timing budgets are wired into the build pipeline, so the next campaign tag or oversized banner fails a check before launch instead of undoing the work three releases later. A regression caught by a machine, not by customers.
The metrics
Chasing a Core Web Vitals score is treating the thermometer
Core Web Vitals are useful the way a temperature reading is useful: they tell you something is wrong and roughly where. They are a poor target, because the tricks that move a score without moving the experience are well known — deferring what the metric watches, gaming the largest paint, shifting work to just after the measurement window. A site can pass its vitals and still lose the visitor who tapped a button that did nothing for two seconds.
We treat the vitals as symptoms and the visitor’s task as the goal. The questions that drive the work are commercial: how long until the page a paid click lands on is usable, how long until the form accepts input, how often the checkout stalls on a mid-range phone over a mobile connection. When those improve, the scores follow as a by-product — and the improvement shows up in metrics the business already tracks, which is the only place speed was ever worth buying.
The causes
What is usually stealing the seconds, and what fixing each returns
Most slow sites are slow for a short list of reasons. The diagnosis tells us which of these your site has; the list explains why each one costs money, not just milliseconds.
| Variable | Effect on price | Why |
|---|---|---|
| Media shipped at capture size | The largest single delay on most marketing pages | Hero images and video exported straight from the design or the camera dwarf everything else on the wire. Sizing, modern formats and lazy loading are unglamorous and routinely worth more than any code change. |
| Third-party tags nobody audits | Slowness that returns after every marketing campaign | Analytics, chat widgets, ad pixels and heat-map scripts accumulate one campaign at a time, each defended by a different team. An audit that removes the dead ones and defers the rest often beats months of front-end tuning. |
| JavaScript that blocks the first interaction | A page that looks ready but ignores the visitor | Oversized bundles keep the main thread busy after the paint, so the visitor sees the page and taps a button that does nothing. Splitting, deferring and deleting are the cures, in that order of difficulty and the reverse order of glamour. |
| Distance and caching left to chance | Every visit pays a geography tax it does not need to | A site served from a single far-away origin, without edge caching, makes every Saudi visitor pay for the distance on every request. Serving cached content from nodes close to the Gulf removes a delay no amount of code optimisation can touch. |
| A backend that thinks before it answers | A floor under every page time no front-end fix can lower | Slow queries, missing indexes and chatty internal calls set the minimum time any page can respond in. When the field data points here, the fix is in the application, and pretending otherwise wastes the front-end budget. |
Ask any supplier quoting this
- Which field metric, on which pages, will this work improve — and how will we see it in our own analytics?
- What stops the site slowing down again after you leave?
Our role
What we diagnose and repair, and the guard your team inherits
- 01
Evidence before invoices
The engagement starts with measurement, and the findings are yours whether or not we do the fixing — a ranked, costed list of causes beats a proposal that promises speed in the abstract.
- 02
Fixes across the whole path
We work wherever the cause lives — media pipeline, script loading, third-party audit, caching and delivery, backend response — rather than stopping at the layer a front-end-only engagement can reach.
- 03
A budget your pipeline enforces
We leave performance budgets running in your build, with your team trained to read them, so the speed becomes a property the organisation defends rather than a snapshot from the month we were there.
Which pages matter most commercially is your call and steers the whole sequence; we supply the measurement, the causes, the fixes and the guard that keeps the next release from spending what this one saved.
Common questions
What buyers ask
Our Lighthouse score is green but customers still complain the site is slow. What is going on?+
Lighthouse is a lab test: one synthetic load, on one simulated device, from wherever the test ran. Your customers load the site on mid-range phones, over mobile networks, with a cart in progress and a dozen third-party scripts fighting for the main thread — conditions the lab run never sees. The two disagree constantly, and when they do, the field is right. This is why we anchor the work in real-visitor measurement: it finds the slowness your customers are describing rather than certifying the version of the site that only the test machine meets.
Can a slow site be made fast without rebuilding it?+
Usually, yes. Most of the causes we find — oversized media, unaudited third-party tags, blocking scripts, absent caching — are repairable on the site you have, and repairing them is far cheaper than replatforming. The honest exception is a backend or a platform whose minimum response time is itself the problem; no front-end work lowers that floor, and we will say so early rather than optimise around it for months. The diagnosis exists precisely to make that call before money is spent, not after.
Does a site serving Saudi visitors need infrastructure inside the region to feel fast?+
It needs its content served from near the region, which is not the same as moving your hosting. A content delivery network with nodes close to the Gulf serves the cacheable majority of the site — pages, images, scripts — from nearby, and only the genuinely dynamic requests travel to the origin. For most sites that removes the distance penalty at a fraction of a migration’s cost. Where the dynamic part is the slow part, regional hosting enters the conversation, but as a measured decision about specific requests, not a reflex.
In this cluster
A performance review that names the slowest commercially important pages, the causes behind them, and the fixes ranked by expected revenue effect.
Request a review
